higher-performance

Ahh, okay I see. I think changing the existing API might be possible? Regardless -- it might be a bit before I get around to this, but I'll get back...

Hi, I took a deeper look at this. It turns out to be much less trivial than this PR suggests, since actually creating a file has additional failure modes (read-only...

Note that you will almost certainly need to new tests for this (in addition to making sure all the existing ones pass, even disabled ones) to test all the edge...

@ppanchad-amd: I'm confused, I still see these in the sources. What do you believe has changed? https://github.com/ROCm/MIOpen/blob/261a49738aba61ce50a77e717895de9619de5f75/test/perfdb.cpp#L942

Ah I see. It's been a while, so at this point I don't recall how I came across it; it's possible I just saw the code and felt it looked...

Hi! I'm a little bit confused, how does moving the global into a function avoid the problem of multiple constructions? Wouldn't each shared library still get its own copy of...

> FWIW I am seeing this same issue when compiling against the googletest source, so I don't think the shared / static conversation has anything to do with it. Not...

Thank you, but I suspect we won't have the bandwidth to to track down such an issue in a massive project like this. I'd really need something minimal that I...

I'm going to close this PR, since I think we need to see something indicating that (a) an issue actually exists inside GoogleTest (as opposed to the issue being that...

@CastleOnTheHill: Fundamentally it doesn't really make sense to have multiple copies of GoogleTest inside a single program to begin with. To give an analogy: it's like designing a car so...